Two New Image Encryption Schemes Based On Differential Privacy Protection And Arnold Transformation

With the rapid development of information society,the relevant application of image information is more and more extensive,especially in the field of artificial intelligence.However,a common image may contain a large amount of personal privacy information,which can not be imagined once the sensitive information is leaked.Therefore,it is necessary to encrypt the digital image.Arnold transform is one of the most common image encryption methods.It is widely used in image encryption because of its simple transformation mode and remarkable scrambling effect.However,Arnold transform also has some limitations,such as the histogram of ciphertext image and clear text image is identical,which is easy to expose statistical law.Therefore,more secure encryption scheme is urgently needed.Differential privacy technology,as a new technology in recent ten years,is widely used in artificial intelligence because of its efficient encryption method.This paper first combines differential privacy technology with image encryption,and proposes two secure and efficient Arnold transform image encryption schemes based on differential privacy protection.The two schemes use the differential privacy noise mechanism to act on the ciphertext image after Arnold transformation,which improves the security and efficiency of image encryption.Firstly,the paper reviews the development of image encryption and the application of differential privacy technology,and analyzes the feasibility of using differential privacy technology in image encryption.Then,the Arnold transform and differential privacy technology are introduced respectively.The focus is on how to combine differential privacy technology and Arnold transform in image encryption.In view of the Arnold transform image encryption scheme based on differential privacy protection,this paper has carried out the following two tasks:(1)The image encryption scheme based on the global disturbance of Laplace noise is designed.By using Laplace mechanism of differential privacy,the global disturbance of ciphertext image after Arnold transform is carried out,which further reduces the correlation between adjacent pixels of ciphertext image,thus effectively resists differential attack and statistical attack.(2)The image encryption scheme based on differential privacy index is designed.By using the differential privacy index mechanism,combined with singular value decomposition of matrix,some elements of singular value matrix are selected,and Laplace noise disturbance is carried out.Finally,the disturbed matrix is reset.The whole process satisfies ?-differential privacy,thus improving the scheme encryption efficiency.Finally,the paper designs two Arnold transform image encryption schemes based on differential privacy protection,which are all implemented by MATLAB programming,which verifies the feasibility of the scheme.At the same time,the data is analyzed and compared from histogram,adjacent pixel correlation,information entropy,nPCR and uaci,so as to evaluate the security and efficiency of the scheme.The results show that the security of ciphertext image is enhanced by combining differential privacy technology with Arnold transform,and Laplace noise mechanism is used to enhance the security of ciphertext image.Some elements of singular value matrix are randomly selected by the index mechanism of differential privacy to balance the availability and privacy of ciphertext image.Therefore,the scheme proposed in this paper has a high application value and practical significance for the construction of image security information storage system.